Mathematical modelling of a basin system

Human life is intertwined with nature. The changes that take place in the nature directly affect the human life. Natural disasters such as droughts and floods have been one of the most important problems faced by people throughout human history. Although these natural disasters have many reasons, one of the most important reasons is that water resources are not used efficiently and well managed. To use water resources efficiently, it is necessary to know the daily and annual flow values of the rivers in the present region and to go to action accordingly. In order to manage water resources and take measures against possible situations, it is very important to estimate the flow values. Artificial intelligence methods are used effectively and successfully in many areas for a long time. Recently, the use of hydrometeorological field has increased. Artificial neural networks, which are successfully used in linear and nonlinear systems and are a black box model, are particularly successful in modeling systems where the physical model is difficult and complex, or whose physical model can not be precisely defined. Artificial neural networks are easy to apply because they do not perform complicated mathematical operations found in physical models and because they model the system with only input-output data. This is a highly preferred method due to its advantages. The prediction success of artificial neural networks varies according to the applied field. The wavelet transform method is used to increase this prediction success. Wavelet transform is a method that analyzes a signal in frequency and time space. In wavelet analysis, both the frequency and the time resolution are good, so the information carried by the mark can be examined better. In this study, daily meteorological and flow data of the Ergene basin scattered stations in Trakya region were chosen to estimate daily flow rate. Meteorological and flow data of 8 stations were selected to estimate. Only the flow rate at the first stage of the work was used to make predictions in the conventional ANN method. Then the same original data was separated into its components by discrete wavelet transform and the most effective components were selected. These selected components are presented as inputs to the ANN and the flow rate is estimated. When the results of these two models are compared and performance criteria are taken into account, it has been observed that the wavelet artificial neural network performs better than the conventional artificial neural network. In the study, both meteorological data and flow data were then used to estimate the flow rate. This data is separated into its components by discrete wavelet transform and the most active components are presented as inputs to the ANN model. The use of meteorological data has increased the prediction success. In this study, it is aimed to estimate the daily flow rate using artificial neural network and wavelet analysis methods by using meteorological data and flow data. It has been shown that artificial neural networks and wavelet transform methods can be successfully used together in the flow estimation.